Transfer RNAs (tRNA) are short non-coding RNAs comprised of 75–95 nucleotides, universally present in all organisms from the prokaryotes to the eukaryotes. The 75–95 residues polynucleotide sequences of tRNAs fold back upon themselves and form hydrogen-bonded clover leaf-like structures that fold into L-shaped tertiary structure (Holley et al., 1965; Wilusz, 2015). The tRNA is organized into a linear double stranded helix. Determination of the three-dimensional structure of a tRNA was a landmark discovery for modern molecular biology research. The major function of the tRNA is to bridge between an amino acid and an mRNA during protein synthesis where the tRNA transfers the cognate amino acid to the translating polypeptide chain. The clover leaf-like structure of the tRNA is comprised of an acceptor arm, D-arm, D-loop, anti-codon arm, anti-codon loop, variable loop, Ψ-arm, and Ψ-loop (Figure 1; Kirchner and Ignatova, 2014; Mohanta and Bae, 2017). The acceptor arm is 7 base pairs long, the D-arm is 3–4 base pairs, the D-loop is 4–12 nucleotides, the anti-codon arm is 5 base pairs, the anti-codon loop is 7 nucleotides, the variable loop is 4–23 nucleotides, the Ψ-arm is 5 base pairs and the Ψ-loop is 7 nucleotides long (Figure 1; Kirchner and Ignatova, 2014). The heterogeneity in the length of tRNAs is caused by variable number of bases in D-loop and the variable loop. The most important functional parts of a tRNA are: the anti-codon triplet, which reads the codons of a messenger RNA; a 3′-CCA tail where the cognate amino acid is attached and the Ψ-arm and Ψ-loop that hold the ribosome machinery.

There are 21 different types of iso-acceptor families for the 20 amino acids, one for each amino acid and one for selenocysteine. In prokaryotes, an isoacceptor family may have one (tRNA^Trp, and tRNA^Met) to six tRNA members (tRNA^Ser, tRNA^Arg, and tRNA^Leu). To date, four different types of tRNA genes have been reported so far including intron-containing, non-intronic, split and permutated types (Randau et al., 2005; Fujishima et al., 2009; Chan et al., 2011). The intron-containing and non-intronic tRNAs are encoded in a single gene, whereas the split tRNAs are encoded in two or more separate genes; in permutated tRNAs, the 3′ half of the tRNA is located up-stream to the 5′ half of the tRNA (Tocchini-Valentini et al., 2009; Maruyama et al., 2010; Soma, 2014). The intron-containing, permutated and split tRNA genes are called “disrupted tRNA genes” whereas the non-intronic tRNA genes are called “continuous tRNA genes” (Sugahara et al., 2009; Kanai, 2015). The disrupted tRNA genes may have evolved earlier than the continuous genes (Di Giulio, 2008a,b; Fujishima et al., 2009). The tRNA mini helix is the most ancient form of tRNA and forms the top half of the tRNA that contains the acceptor arm and the Ψ-arm (Weiner and Maizels, 1987, 1999; Shi et al., 1992, 1998; Sun and Caetano-Anollés, 2008). The tRNA mini-helix is a validated substrate for aminoacyl-tRNA synthetases and the 3′-CCA tail adding enzymes (Shi et al., 1992, 1998). The genomic tag hypothesis suggests that the upper half of the tRNA had evolved earlier than the lower half. Phylogenetic analysis also suggests that the top half of the tRNA is more ancient than the bottom half (Sun and Caetano-Anollés, 2008). The evolution of tRNA occurred with respect to its surrounding environment and is far more complex than that of proteins, because tRNA perform multiple roles in the cell (Navarre and Schneewind, 1999; Kirchner and Ignatova, 2014). tRNA interacts with numerous molecules; during protein synthesis, it interacts with amino-acyl tRNA synthetase, ribosome and messenger RNA, and a large number of transcription factors and other enzymes (Nirenberg and Leder, 1964). tRNA^Gly acts as a structural element of the peptidoglycan of the bacterial cell wall whereas tRNA^Lys and tRNA^Ala are used to aminoacylate membrane lipid to permeate cationic antibiotics (Roy and Ibba, 2008). Cyanobacteria are the most ancient living organism whose existence dates back to 3.3–3.5 billion years (Schopf and Packer, 1987; Altermann et al., 2006). Because it is most likely that the tRNA genes date back to the age of cyanobacteria, and studying the genomic and evolutionary perspectives of cyanobacterial tRNA is very important. Given their high level of conservation it is especially important to study the genomics, origin and evolution of tRNAs. Here, a comprehensive study of cyanobacterial tRNA was conducted by analyzing the tRNA sequences of 61 cyanobacterial species.

Nucleotide sequence conservation is an important phenomenon that demonstrates conserved functional role. Therefore, understanding the conserved nucleotide consensus sequences in cyanobacterial tRNA was very important. In this study we found that except for tRNA^Asn, tRNA^Leu, tRNA^Gln, and tRNA^Met, all the tRNAs were contained conserved G¹ nucleotide at the 1st position in the acceptor arm. tRNA^Leu and tRNA^Met did not contain any conserved nucleotides at the 1st position whereas tRNA^Asn and tRNA^Gln contained a conserved U¹ nucleotide at the 1st position. With a few exceptions, the G² nucleotide was conserved at the 2nd position in tRNA^Ala, tRNA^Val, tRNA^Ile, tRNA^Phe, tRNA^Tyr, tRNA^Ser, tRNA^Gln, tRNA^Lys, tRNA^His, tRNA^Arg, and tRNA^Asp. The nucleotide C² was conserved in tRNA^Gly and tRNA^Thr whereas tRNA^His contained either a C² or a U² nucleotide. At the 3rd position, the G³ nucleotide was conserved in tRNA^Gly, tRNA^Ala, tRNA^Pro, tRNA^Ile, tRNA^Tyr, tRNA^Lys, and tRNA^Asp, whereas tRNA^Phe contained a C³; tRNA^His contained C³/U³; whereas tRNA^Ser contained an A³ nucleotide. At the 4th position of the acceptor arm, tRNA^Val and tRNA^Ile contained the conserved C⁴ nucleotide whereas tRNA^Tyr contained a U⁴ nucleotide and tRNA^Pro contained the G⁴ nucleotide. In tRNA, the C⁵ nucleotide was conserved, whereas tRNA^Phe and tRNA^Gln contained the conserved G⁶ nucleotide. In all canonical tRNA, 1st, 2nd, and 3rd base of the acceptor arm pairs with the 72nd, 71st, and 70th base, respectively of the acceptor arm. Their conserved structure is highly important to have proper functional clover leaf-like structure of tRNA. At the 7th position, tRNA^Pro and tRNA^His contained conserved G⁷ nucleotide. In tRNA^Tyr, all the nucleotides in the positions 1–5 were conserved across the species. In the 1st and 2nd positions of the acceptor arm of tRNA, the G¹-G² conserved consensus sequence present in tRNA^Ala, tRNA^Val, tRNA^Ile, tRNA^Tyr, tRNA^Ser, tRNA^Lys, tRNA^Arg, and tRNA^Asp whereas the G¹-C² conserved consensus sequence present in tRNA^Gly, tRNA^Phe, tRNA^Thr, and tRNA^Lys. The nucleotide, at the 2nd position of the acceptor arm, was substituted with either G² or C², whereas only tRNA^His had a U² at the 2nd position. Only tRNA^Asn and tRNA^Gln contained a U¹ nucleotide at the 1st position of the acceptor arm whereas tRNA^Tyr contained conserved U⁴ nucleotide at the 4th position and tRNA^His contained the conserved U² nucleotide at the 2nd position (Table 2). Except the above-mentioned four tRNAs, none of the other tRNAs were found to encode U nucleotide in the acceptor arm. Previously, the 8th position of tRNA between the acceptor arm and the D-arm was thought to contain a conserved U⁸ nucleotide (Table 3). In this study, the majority of tRNAs were found to contain a conserved U⁸ nucleotide. However, tRNA^His was found to contain either a U⁸ or an A⁸ whereas tRNA^Glu contained a U⁸, C⁸ or a G⁸ nucleotide (Table 3). This showed that the 8th position of cyanobacterial tRNA can be substituted with any nucleotide. Similarly, at the 9th position, the majority of tRNAs were found to contain a conserved A⁹ nucleotide (Table 3). However, tRNA^Leu, tRNA^Tyr and tRNA^Ser were found to contain a G⁹ nucleotide whereas tRNA^Thr, tRNA^Lys, and tRNA^Arg contained an A⁹ or a G⁹ nucleotide (Table 3). tRNA^His contained an A⁹ or a C⁹ nucleotide at the 9th position whereas tRNA^Cys and tRNA^Gln contained either a A⁹, C⁹, or a G⁹ (Table 3). tRNA^Met contained either an A⁹, G⁹ or a U⁹ nucleotide whereas tRNA^Glu contained an A⁹, C⁹, or a U⁹ nucleotide at the 9th position. This negated the previous assumption about the conservation of the U⁸ nucleotide in cyanobacterial tRNA and showed that it can be prone to variations; the major variation were observed in tRNA^Glu which was found to contain varied nucleotides at 8th and 9th position.

In the D-arm, the majority of tRNA contained a conserved G¹ nucleotide at the 1st position except for tRNA^Gly, tRNA^Pro, tRNA^Val, tRNA^Met, tRNA^Gln, tRNA^Lys, and tRNA^Glu. In total of 20 tRNA families, 10 families were found to contain a conserved C² nucleotide at the 2nd position (Table 2). These were tRNA^Ala, tRNA^Val, tRNA^Ile, tRNA^Phe, tRNA^Ser, tRNA^Thr, tRNA^Cys, tRNA^Asn, tRNA^Lys, and tRNA^His whereas tRNA^Met was found to contain a G² and tRNA^Trp contained a U² nucleotide at the 2nd position (Table 2). With respect to the first two nucleotides in the D-arm, the G¹-C² conserved consensus sequence can be found in tRNA^Ala, tRNA^Ile, tRNA^Phe, tRNA^Ser, tRNA^Thr, tRNA^Cys, tRNA^Asn, and tRNA^His. At the 3rd position, the U³ nucleotide conserved in tRNA^Ala, tRNA^Ile, tRNA^Phe, tRNA^Asn, and tRNA^Asp. At the 4th position of the D-arm, the C⁴ nucleotide was found to be conserved in tRNA^Ala, tRNA^Val, tRNA^Ile, tRNA^Phe, tRNA^Tyr, tRNA^Thr, and tRNA^Asp (Table 2). The D-stem expected to have rigid structure due to high G/C content, but due to the presence of more A-U and G-U base pairing, it seems rather weak (Hardt et al., 1993). Previous study reported that D-arm play important role in recognition of aminoacyl-tRNA synthetase and is highly variable and possess unusual conformation due to over-crowding of G residue (Smith and Yarus, 1989; Hardt et al., 1993). However, in our study we found that D-arm is also highly conserved, suggesting its common functional role.

Except for tRNA^Met and tRNA^Glu, 18 tRNA families were found to contain a conserved A¹ nucleotide at the 1st position in the D-loop. At the 2nd position, the G² nucleotide was conserved in tRNA^Pro, tRNA^Val, tRNA^Phe, and tRNA^Tyr, whereas tRNA^Thr contained a U² nucleotide at the 2nd position (Table 2). In the 3rd position, the U³ nucleotide was conserved in tRNA^Phe, and tRNA^Tyr and the G³ nucleotide was conserved in tRNA^Cys, tRNA^Asn, and tRNA^His. In the 4th position of the D-loop, the G⁴ nucleotide was conserved in tRNA^Gly, tRNA^Tyr, tRNA^Thr, tRNA^Asn, tRNA^Gln, tRNA^Lys, tRNA^Arg, and tRNA^Asp whereas at the same position tRNA^Phe contained conserved U⁴ nucleotide (Table 2). At the 5th position, the G⁵ nucleotide was found to be conserved in tRNA^Gly, tRNA^Ala, tRNA^Leu, tRNA^Phe, tRNA^Tyr, tRNA^Trp, tRNA^Ser, tRNA^Cys, tRNA^Lys, tRNA^His, tRNA^Asp and tRNA^Glu. The majority of tRNAs were found to contain a conserved G⁵ nucleotide whereas tRNA^Thr contain a U⁵ nucleotide. At the 6th position, tRNA^Ala, tRNA^Val, tRNA^Leu, tRNA^Ile, tRNA^Met, tRNA^Phe, tRNA^Ser, tRNA^Cys, and tRNA^His contained the G⁶ nucleotide whereas tRNA^Tyr, and tRNA^Asp contained the U⁶ nucleotide. At the 7th position of the D-loop, the U⁷ nucleotide was found to be conserved in tRNA^Phe, tRNA^Tyr, tRNA^Cys, and tRNA^His whereas the 8th position had a conserved A⁸ nucleotide in tRNA^Ala, tRNA^Leu, tRNA^Tyr, tRNA^Arg and tRNA^Asp. The A⁹ nucleotide was conserved in tRNA^Val, and tRNA^Ile. This indicates that a great variation was present in the nucleotide composition of the D-loop. However, the A¹ nucleotide was conserved in the majority of the cases, whereas conservation of nucleotides in the other parts of the D-loop was specific to an individual tRNA family. The D-loop interacts with the Ψ-loop via long range interactions where G¹⁸ and G¹⁹ interact with Ψ⁵⁵ and C⁵⁶, respectively (Hanawa-Suetsugu et al., 2001). Although the position is dynamic, except for tRNA^Gln, all tRNAs were contained at least one conserved G nucleotide in the D-loop, suggesting their common functional role.

The anti-codon arm of tRNA was less conserved than other parts. In tRNA^Phe, tRNA^Trp and tRNA^His, the G⁴ nucleotide was found to be conserved, whereas tRNA^Trp had a conserved U⁵ nucleotide and in tRNA^His, the G⁵ nucleotide was conserved. The anti-codon loop, which reads the codon of an mRNA during protein translation, is the most important part of tRNA. The C¹ nucleotide was found to be conserved at the 1st position of the anti-codon loop in tRNA^Ile, tRNA^Met, tRNA^Trp, tRNA^Cys, tRNA^Asn, tRNA^Lys, and tRNA^Asp. Additionally, the U¹ nucleotide at the 1st position of the anti-codon loop, was found to be conserved in tRNA^Pro, tRNA^Val, tRNA^Gln, tRNA^His, and tRNA^Arg (Table 2). Occasionally, tRNA^Arg contained a C¹ nucleotide in the anti-codon loop. All the tRNAs were found to contain a conserved U² nucleotide at the 2nd position of the anti-codon loop. The A⁶ nucleotide, at the 6th position of the anti-codon loop, was conserved in tRNA^Gly, tRNA^Ala, tRNA^Val, tRNA^Ile, tRNA^Met, tRNA^Trp, tRNA^Ser, tRNA^Thr, tRNA^Cys, tRNA^Asn, tRNA^Lys, and tRNA^Asp (Table 2). In the 7th position of the anti-codon loop, the A⁷ nucleotide was conserved in tRNA^Ile, tRNA^Met, tRNA^Trp, tRNA^Thr, tRNA^Cys, tRNA^Asn, and tRNA^Lys. The presence of a conserved A⁶ nucleotide at the 6th position suggests the possible entry site for group I intron in the anti-codon loop.

Previous study reported that the variable loop of tRNA is dynamic in nature and does not contain any conserved nucleotides. However, in our study we found that the variable loop contained conserved nucleotides at the specific positions at the level of individual tRNA family (Table 2). tRNA^Val and tRNA^Phe were found to contain the conserved G³-U⁴-C⁵ nucleotides at the 3rd, 4th, and 5th positions of the variable loop, whereas tRNA^Pro contained the conserved G³-x-C⁵ nucleotides at 3rd and 5th position (Table 2). In tRNA^Ile and tRNA^Asn, the G³-U⁴ nucleotides were found to be conserved at 3rd and 4th positions. At the 5th position, the C⁵ nucleotide was conserved in tRNA^Pro, tRNA^Val, tRNA^Phe, and tRNA^Cys (Table 2). This study indicates that the G³, U⁴, and C⁵ nucleotides at 3rd, 4th, and 5th position, respectively, were the most likely to be conserved in the variable loop of the cyanobacterial tRNA.

All the tRNAs were found to contain a conserved G⁵ nucleotide in the Ψ-arm. Several tRNAs contained a conserved G⁴ nucleotide at the 4th position whereas others was found to contain conserved G³ nucleotide. The conserved G nucleotide at the 4th and 5th position was detected as the G⁴-G⁵ configuration in tRNA^Pro, tRNA^Val, tRNA^Ile, tRNA^Tyr, tRNA^Ser, tRNA^Gln, tRNA^Lys, tRNA^His, and tRNA^Arg.

The Ψ-loop of all the tRNA contained the conserved U¹-U²-C³-x-A⁵ nucleotides at the 1st, 2nd, 3rd, and 5th positions, respectively. Few other tRNAs were found to contain the extended conserved U¹-U²-C³-x-A⁵-x-U⁷ sequence from the 1st to the 7th position of the Ψ-loop (Table 2). The tRNA containing the conserved U¹-U²-C³-x-A⁵-x-U⁷ nucleotides were tRNA^Leu, tRNA^Ile, tRNA^Met, tRNA^Tyr, tRNA^Trp, tRNA^Ser, tRNA^Cys, tRNA^Asn, tRNA^Lys, tRNA^His, tRNA^Arg, and tRNA^Glu (Table 2). tRNA^Pro was found to contain the conserved U¹-U²-C³-x-A⁵-A⁶-U⁷ consensus sequence, where the A⁶ nucleotide was found to be conserved at the 6th position. In tRNA^Asp, it was found to contain conserved U¹-U²-C³-x-A⁵-x-C⁷ consensus sequence in which the U⁷ nucleotide substituted with a C⁷ nucleotide (Table 2). These analyses show that tRNAs were conserved in the acceptor arm, D-arm, D-loop, anti-codon arm, anti-codon loop, the variable loop and the Ψ-arm at the individual family level with majority of tRNAs showing greater conservation in the Ψ-loop.

In spite of the presence of highly conserved genetic architecture of cyaobacterial tRNA, they also found to possess introns within it. Previous studies reported that an intron was present only in tRNA^Leu (UAA) and tRNA^fMet (UAC) of cyanobacterial tRNA (Paquin et al., 1997; Rudi and Jakobsen, 1999). However in our study, we found that the cyanobacterial group I intron was also present in tRNA^Arg, tRNA^Gly, and tRNA^Lys (Figure 3). Because cyanobacteria are a prokaryotic organism, the intron, present in cyanobacterial tRNA can be compared with the group I intron in the plastid. Previous studies reported that the group I intron, found in the plastid, are immobile (Rudi and Jakobsen, 1999). The group I introns found in various other locations including the mitochondria, nucleus, bacteria, and bacteriophages share conserved sequence motifs and have evolved via lateral gene transfer (Rudi and Jakobsen, 1999). Cyanobacterial group I introns were also possessed conserved motifs. Rudi and Jakobsen (1999) have classified the group I intron as intron element P, Q, R, and S. Cluster analysis revealed that tRNA^Arg of the Nostoc sp. PCC 7524 were grouped in intron element P, whereas the introns of the Nostoc sp. PCC 7107 and Gloeocapsa sp. PCC 73106 were clustered independently. None of these were clustered with intron elements Q, R, or S. This suggests that there may be another conserved intron element in the cyanobacteria which is yet to be elucidated. The presence of an intron, in the anti-codon loop of cyanobacterial tRNA agrees with the exon theory of the evolution of tRNA genes where modern tRNA is reported to have evolved by duplication of one half of tRNA (Di Giulio, 1992). Report suggests that the introns of cyanobacterial tRNA moved via lateral gene transfer (Marck and Grosjean, 2002).

Self-splicing group I intron of cyanobacterial tRNA^fMet in genera Dermocarpa, Scytonema, and Synechocystis possess open reading frame (ORF). The intron of Synechocystis possesses ORF of 150 codons. The introns in cyanobacteria distributed sporadically without any correlation between relatedness in intron sequence corroborates with the lateral transfer of intron (Dujon, 1989; Biniszkiewicz et al., 1994). Sporadic occurrence of these group I introns in homologous genes of closely related species show consistent similarity in the mobility. The group I introns are suggested to be either mobile and ancient in origin (Rudi and Jakobsen, 1999). Closely related group I intron are inserted in the UAA anit-codon of tRNA^Leu of cyanobacteria and chloroplast (Biniszkiewicz et al., 1994). Study of introns from 22 cyanobacterial strains by Rudi and Jakobsen (1999) demonstrated relatively recent gain and/or loss of intron in some cyanobacterial lineage (Rudi and Jakobsen, 1999). The presence of self-splicing group I intron in tRNA^Leu at the same position of the same gene in cyanobacteria and chloroplast indicated that this intron predates the invasion of eukaryotic cells by endosymbiosis (Reinhold-Hurek and Shub, 1992). The group I intron of tRNA^Leu have higher homology with the group I intron of tRNA^Ile (CAU) and tRNA^Arg (CCU) (Rudi and Jakobsen, 1997, 1999).

Evolution of a gene occurred through random mutation/substitution. The pattern and frequencies of nucleotide substitution is largely depends on the mutational events of the gene (Shimizu et al., 1989; Zhang and Gerstein, 2003; Arnheim and Calabrese, 2009). Although transitions are more frequent in coding genes because they are less likely to result in amino acid substitutions due to a wobble in the genetic code, the presence of a higher rate of transition in cyanobacterial tRNA is intriguing. A universal bias may favor transition over transversion and a diverse rate of transitions/transversions, combined with the bias may help to explain the multiple evolutionary rates and lineages. The rates of transition/transversion bias of tRNA^Trp, tRNA^Asp, and tRNA^Tyr were 8.409, 4.131, and 3.658, respectively; these were quite higher than the rates of other tRNAs (Table 6). At a low level of genetic divergence, the transition/transversions bias remains high, whereas at a high level of genetic divergence, the transition/transversion bias remains low (Yang and Yoder, 1999). This indicates that tRNA^Trp, tRNA^Asp, and tRNA^Tyr have a very low level of genetic divergence, whereas other tRNAs have a very high rate of genetic divergence. Therefore, cyanobacterial species have only a fewer number of tRNA^Trp, tRNA^Asp, and tRNA^Tyr genes in their genome compared to other tRNAs and rest of the tRNAs have a very high rate of divergence. Due to high genetic divergence, the frequency of other tRNAs are more abundant than tRNA^Trp, tRNA^Asp, and tRNA^Tyr. Silent mutation can have significant impact on exhibiting genetic divergence which accompany through selection/genetic drift or novel adaptation (Palumbi, 1994).

Understanding the process of evolution and function of new gene is important in genomics and evolutionary biology; gene duplication is a powerful process in generating novel gene function via sub-functionalization and neo-functionalization, whereas loss of a gene can greatly shape the gene family (Lynch and Conery, 2000; Niimura and Nei, 2007; Rasmussen and Kellis, 2012; Teufel et al., 2016). Gene duplication is a fundamental process in the evolution of novel species, especially in eukaryotes in which it plays major roles in creating novel gene functions (Cotton and Page, 2005; Blomme et al., 2006; Zhao et al., 2015). Similarly, the loss of a gene occurs via segmental deletion, and pseudogenization preserves a minimum number of functional gene copies (Ohno, 1970; Lynch and Conery, 2000; Ting et al., 2004; Cotton and Page, 2005; Blomme et al., 2006; Demuth et al., 2006). Due to the lowest percentage of duplication, the genome of the cyanobacteria encodes only one to two tRNA^His genes per genome, whereas the number of tRNA^Ile genes is more numerous due to highest percentage of duplication (Table 5). The low level of duplication and conditional duplication may responsible for the low frequency in the occurrence of tRNA^His, tRNA^Asn, and tRNA^Asp in the cyanobacterial genome. Coupled with the low percentage of duplication, conditional duplication and the high percentage of gene loss led to the low frequency of tRNA^His, tRNA^Asn, tRNA^Gln, and tRNA^Asp genes in the cyanobacterial genome. Similarly, low percentage of gene loss, coupled with the high percentage of gene duplication and conditional duplication in tRNA^Arg and tRNA^Ala, resulted multiple and variable copies of tRNA^Arg and tRNA^Ala in the cyanobacterial genome (Table 5). Previous studies showed that tRNAs are the classic example of gene families that were continuously duplicated and lost (Withers et al., 2006; Tang et al., 2009; Bermudez-Santana et al., 2010; Rogers et al., 2010). The operon or tRNA clusters are highly dynamic and unstable genomic region. Therefore, it became an easy target for duplication or loss event. However, in cyanobacteria, the loss event was predominant over the duplication and conditional duplication.

Numerous studies have been conducted on the evolution of tRNA. Reports suggest that tRNA hairpin structures were the precursors of tRNA, having evolved from the combination of different sub-structures, where the top half (acceptor arm) of the tRNA is considered older than the bottom half containing the anti-codon arm (Sun and Caetano-Anollés, 2008). According to the “genomic tag” hypothesis, the upper half of the tRNA harbors the ancestral genomic information and the bottom half provides the specificity for the genetic code (Weiner and Maizels, 1987). The hairpin structure having the anti-codon loop may have evolved as an intermediary in the process of protein synthesis. Introns in tRNA are usually present in the anti-codon loop, and the 5′ and 3′ halves of the tRNA were likely the mini-genes before their merger (Di Giulio, 1992, 2004). The presence of an intron in tRNA suggests that the two modules of tRNA were separated by introns; hence, tRNA is the exon part of the gene described by the “exonic gene theory” of the origin of tRNA (Di Giulio, 1995). The tRNA molecule is universally present in all cellular organisms and is transferred by “horizontal gene transfer.” It's ubiquitous presence in all cellular lineages strongly supports its early origin, explaining its presence in organisms of early origin. Fossil record suggests that cyanobacteria are older than 2 billion years and indirect evidence suggests that they may be older than 2.45 billion years (Amard and Bertrand-Sarfati, 1997; Bekker et al., 2004; Schirrmeister et al., 2011; Mohanta et al., 2017). The presence of tRNA^Arg and tRNA^Met in clusters I, III, V, and VI in the phylogenetic tree reflected their ancient origin (Figure 5); hence, their co-distribution in different clusters may have been followed by subsequent evolution of tRNA^Gln and tRNA^Ala. The presence of tRNA^Arg and tRNA^Met in multiple clusters vividly illustrates that they have undergone duplication and subsequently diverged into different species. The distribution of tRNA^Ile, tRNA^Glu, tRNA^Asp, tRNA^pro, tRNA^Thr, tRNA^Gly, tRNA^Trp, tRNA^His, tRNA^Cys, tRNA^Tyr, and tRNA^Ser was confined to a single cluster only (Figure 5). This suggests that they have evolved recently, compared with tRNA^Met, tRNA^Arg, tRNA^Ala, and tRNA^Gln (Figure 5). In cluster I, tRNA^Arg grouped twice and present close to tRNA^Asp and tRNA^Gly, suggesting their evolution from tRNA^Arg (Figure 5). In cluster I, tRNA^Ala was present close to tRNA^Val; this suggests that tRNA^Val evolved from tRNA^Ala. In cluster II, tRNA^pro was grouped separately whereas tRNA^Thr, tRNA^Phe, and tRNA^Asn were grouped with tRNA^Met, suggesting their evolution from tRNA^Ala (Figure 5). Cluster III contained tRNA^Arg and tRNA^Trp, suggesting possible evolution of tRNA^Trp from tRNA^Arg. In cluster IV, tRNA^Gln present close to tRNA^His, and tRNA^Met present closer to tRNA^Tyr and tRNA^Cys. This suggests that tRNA^His may have evolved from tRNA^Gln whereas tRNA^Tyr and tRNA^Cys may have evolved from tRNA^Met. tRNA^Ser found in cluster IV shared a branch with tRNA^Met, suggesting the evolution of tRNA^Ser from tRNA^Met. tRNA^Gln, tRNA^Met, and tRNA^Ala were found together in cluster V, suggesting that these tRNAs may have coevolved together, and later diversified and evolved into other tRNAs. These tRNAs were grouped with several pseudo-tRNAs in the cluster of tRNA^Asn (Figure 5, Supplementary Figure 1). This suggests that these tRNAs are likely in the verge of becoming pseudo-tRNAs and will be lost from the evolutionary lineage. tRNA^Phe was present as sister branch to tRNA^Asn in cluster III, suggesting their evolution from tRNA^Asn. Several tRNAs^Lys, tRNAs^His, and tRNAs^Ala in cluster IV were grouped with the pseudo-tRNAs, suggesting their possible loss of function and, hence, extinction from the evolutionary lineage by pseudogenization. However, the evolutionary history of cyanobacterial tRNAs may be more complicated. The presence of common secondary and tertiary structures occurring with an invariant number of nucleotide residues, 20 invariant gene families, and similar functional roles during the translation is the phenomena that are complicated to delineate. It is assumed that for a total of 20 tRNA gene families, there would be 20 branches, corresponding to a group in each tRNA gene family, in the phylogenetic tree. However, this assumption is incorrect. The differences between tRNA families, with regard to their respective amino acid and tRNA specificity were found overlap significantly. The non-isoacceptor tRNAs shared the phylogenetic groups with the isoacceptors, suggesting their evolution from multiple lineages. This shows that tRNAs had evolved by multiple duplication with extensive genetic mutation and subsequent loss of old genes.

Sequence analysis revealed that cyanobacterial tRNAs were dynamic and transiently conserved within their respective tRNA families. However, the U¹-U²-C³ conserved consensus sequence in the Ψ-loop was found in all of the studied tRNAs. Several cyanobacterial tRNAs contained a group I intron in the anti-codon region; further study is required to discover more introns and to understand their specific genomic and evolutionary aspects. The rate of mutational transition of cyanobacterial tRNAs was higher than that of transversion; cyanobacterial tRNA may have evolved polyphyletically by loss of tRNA genes.

TM: Conceived the idea, performed the study, analyzed data, drafted and revised the manuscript. AS: Revised the manuscript. FA: Revised the manuscript. HB: Revised the manuscript.